Skip to main content
← BackOur philosophy

A chat forgets.
A canvas remembers.

Plooi exists because the best design thinking doesn't happen in a scrolling conversation — it happens on a surface you can see, branch and return to. We took the methods practitioners already trust, gave them AI muscle, and laid them out as a canvas your brain was actually built for.

Why we build this

Design methods, within reach.

Plooi is a tool, but the point was never the tool. Design methods are how good products and services get made — and for most of the people making things, they sit out of reach: unwritten, untaught, or locked inside an agency. That gap is what we build against.

Paywallpriced out Degreecredentialed out Studio doorshut out who can use them today everyone who builds
Vision

A better world, by design — where more of what gets built is genuinely designed, because the methods behind it are no longer behind a paywall, a degree or a studio door.

The world we want. Plooi doesn't appear in it — if it did, this would be positioning, not a vision.

Mission

Make design methods accessible to everyone who builds — so better products and services come from the craft, not just from the tool.

What we do about it. This one is ours to be measured against.

Three values · no shortcuts

Built on three things we refuse to compromise.

None of them are about AI doing more. All three are about you doing more with AI — on a surface made for judgment, not just generation.

Craft

You make it well

We refuse the boilerplate AI soup. Every method in Plooi was hand-built from research literature, tested with practitioners, and tuned to support thinking — not replace it.

Clarity

You see every step

Every fold is visible. Every transformation is auditable. No black-box outputs — you can trace any insight back through the methods that produced it. Glass-box reasoning, not magic.

Curation

You decide what stays

AI proposes; you decide. Plooi generates variations, alternatives and refinements — then waits for your call on which to keep. Human-in-the-loop is non-negotiable.

The foundation

Spatial · Agentic · Glass-Box.

Three values need an architecture. Ours rests on three commitments — the reason a Plooi canvas feels different from a prompt box the moment you touch it.

01

Spatial

Work lives in 2D space, not a 1D thread. Position carries meaning, related methods sit together, and the whole shape of your thinking stays in view — pannable, zoomable, never scrolled away.

02

Agentic

Methods aren't static templates — they're agents. Each one takes inputs, does real work, and hands structured output to the next. Chain them and the canvas runs the workflow, not just stores it.

03

Glass-Box

Every step shows its work. Inputs, reasoning and sources are all on the surface, so you can audit an insight, correct a wrong turn, and trust what you ship. No magic, no hidden prompts.

Method taxonomy

Every method has a job. Four categories cover it.

The Plooi method library splits into four functional categories. Mix them like Lego — most workflows use one or two of each.

01Category

Data & Input

Where raw stuff enters
  • Source · PDFs, URLs, CSV
  • Persona · captured archetype
  • Goal · what to optimize for
02Category

Processor & Reasoning

The thinking layer
  • LLM · GPT-5, Claude, Llama
  • Chain-of-Thought · visible reasoning
  • Critic · adversarial check
03Category

Action & Tool

Side effects on the world
  • Search · web research
  • Generative UI · text → wireframe
  • Simulator · scenario testing
04Category

Logic & Control

The conductor
  • Router · conditional flow
  • Aggregator · gather + sort
  • Human-Approval · the pause
Spatial zoom · three altitudes

One canvas. Three altitudes.

The canvas zooms semantically. Pull back for the whole shape of the work; lean in for one cluster of methods; lean further to read a single method's prompt. Same surface, three levels of detail.

METHODEmpathy METHODHMW METHODCrazy 8s AGENTPersona AGENTCritic
AGENT · PERSONA Sarah, 34 PROMPT You are Sarah, a power user. React to {{ wireframe.frame_3 }} MODEL Claude 4.7 CONFIDENCE ● 92% OUTPUT "It's pretty? I guess?"

100% · see the shapeWhole-workflow shape. You read the topology — branches, fan-outs, dead-ends — without reading a single label.

200% · read the labelsCluster scale. Names are visible. You decide which branch to dive into next.

400% · read the promptInside a single method — prompt, model, confidence, output. Full glass-box, nothing hidden.

Anatomy of a pleat

Three kinds of connection. Each tells a story.

Connections in Plooi aren't just lines — they carry meaning. A solid line says "this data flows directly here." A dotted line says "this influences that conceptually." A curved-back line says "iterate until it's good enough."

METHODEmpathy AGENTSynthesizer data
Hard link

Direct transfer

Output becomes input

The output of A becomes the input of B. Strict type-check, no ambiguity. The bread-and-butter of any workflow.

e.g. Empathy Map → Insight Synthesizer
ZONEBrand AGENTCritic influence
Semantic link

Conceptual pull

Shared context

B is influenced by A but doesn't consume its raw output. Used for shared context, themes, or system-prompt inheritance.

e.g. Brand guidelines ⟿ all Critic agents
Gen Critic
Feedback loop

Iterate to good

Generate → critique → repeat

The output keeps refining until the Critic agent gives a green light. Generate, critique, generate again — automatically.

e.g. Concept ⟳ Critic until approved
Composition patterns

Four patterns. Infinite workflows.

Every Plooi canvas — no matter how big — is made of the same four primitive shapes. Once you can read them, you can build anything.

ABC
Pattern 01

Chaining

One method feeds the next. The simplest discipline: A produces, B refines, C delivers. Most workflows start here.

A?BC
Pattern 02

Routing

Pick the right next step based on the input. Negative sentiment? Route to escalation. Positive? File it and move on.

A
Pattern 03

Parallelization

One brief, five concepts at once. Fan out the work, let agents run in parallel, then gather and rank the results.

GenCritic
Pattern 04

Evaluator–Optimizer Loop

Generate → critique → generate again. The output keeps refining until the Critic agent gives a green light.

Confidence states

Trust, by the colour. No hidden hallucinations.

Every method wears its confidence on its border. Green for high, amber for medium, orange for low, terracotta for shaky. Locked methods pause the workflow until a human signs off — so you always know when to trust, when to read, and when to step in.

Method92%

High

92%
Ship it.
Method78%

Medium

78%
Worth a glance.
Method51%

Low

51%
Read carefully.
Method23%

Very low

23%
Re-prompt or replace.
Methodpaused

Awaiting review

Human pending.
Methodgated

Locked gate

Must approve.
The glass-box principle

Open any method. See the seams.

Every Plooi method is a glass box. Open it to see the prompt it's running, which model is wired to it, its confidence score, and exactly where the output came from. No black-box magic — just legible craft.

  • Visual provenance. Trace every output back to its source quote.
  • Confidence-aware. Low-confidence methods wear an amber dashed border.
  • Human gates. Mark any method "requires approval" — the canvas pauses.
Agent · Reasoning
Persona simulator
● 92% confident⏵ running
1 · Prompt
You are Sarah, 34, a power user of {{ product.name }}. React to this onboarding screen as yourself: {{ wireframe.frame_3 }}
2 · Model
Claude 4.7 Sonnet
temp · 0.5·top-p · 0.9swap
3 · Output
"It's pretty? I guess?" I'd want to skip the onboarding tour — I've used this product before. The big "Welcome back!" pop-up is in my way.
↑ traced from interview transcript· session #84 · L 23–28
Interview transcriptTagged quotePersona attributeSimulated reaction
Live traceHuman-in-loop ↳ click to inspect chain
Sub-agents in parallel

One orchestrator. Three minds at once.

A service-design example. The Orchestrator spins up three sub-agents — User, Backend, Support — to simulate a "lost credit card" journey in real time. Their messages stream into one timeline, so you can see exactly where the failure modes happen.

Orchestrator
Lost-card simulator
User Agent
Sarah, 34 · stressed
+0s
I just lost my card. I need to block it. Now.
+18s
Where is the "block card" button? The app is so confusing…
+1m 24s
(switches to phone, dials support)
Backend Agent
API + DB simulator
+0s
GET /api/cards/me → 200 OK · 47ms
+18s
POST /cards/…/block → 503 · DB locked
+1m 24s
Fallback path: customer service ticket auto-created
Support Agent
Tier 1 rep + tooling
+0s
(idle — no incoming ticket)
+18s
(idle)
+1m 24s
Ticket #4928 received. Customer on hold 2m 14s.
⚠︎

Friction surfaced: User waited 1m 6s for the block-card action to fail before discovering the phone fallback. Add: instant push-notification reassurance after 5s of locked-DB retry.

A Plooi grows up

A workflow you keep folding tighter.

A Plooi isn't built once and frozen. It starts as a rough sketch, earns a Critic and a feedback loop, branches for edge cases, and ends up team-owned and re-runnable. The same canvas, six months apart.

First sketch day 0
01 · First sketchday 0

A few empty methods, dropped on a blank sheet. No wires yet — just the shape of an intention.

02 · First runday 1

Filled with a brief and one set of sources. Six methods wired into a first pass. Output: rough, but real.

03 · 10th runmonth 1

You added a Critic, a feedback loop, and a branch for edge cases. Output: now trusted.

04 · Maturemonth 6

Team-owned, branched per project type, exports to JSON for CI. Now genuinely reusable.

The whole point
A canvas, not a chatbox

Start with a sheet.
End with a workflow.

The philosophy only matters if you can feel it. The fastest way is to open a sheet and fold a workflow yourself.